Toward experiment-driven and AI-enhanced driving behavioral research

i

This article has been updated. A Correction to this article was published on .

i

The Original Article was published on

i

The Original Article was published on

i

This article has been updated

i

This article was retracted on

i

This article has been updated

Abstract

Artificial intelligence (AI) is reshaping driving-behavior research, but its value depends on whether data-driven models can be grounded in experimentally validated behavioral mechanisms. This Comment narrows the discussion to a specific mechanism-to-AI question: how can empirical regularities obtained from controlled traffic experiments be translated into constraints, priors, calibration targets, and validation criteria for AI-enhanced behavioral modeling? We synthesize key experimental mechanisms of longitudinal driving behavior, including concave oscillation growth, critical-velocity regimes, spacing-insensitive adaptation, intrinsic heterogeneity, and mean-reverting stochasticity. We then discuss how these mechanisms can guide foundation-model reasoning, digital-twin calibration, federated behavioral aggregation, and safety- or causality-oriented validation. To make this connection explicit, we propose a Hybrid AI–Experimental Behavioral Stack (HBS), not as a generic software architecture, but as a mechanism-to-implementation roadmap linking representation, scenario, policy, transmission, and validation layers. The discussion also clarifies the limitations of foundation models and large language models, including hallucination, weak physical grounding, real-time reliability, and safety verification. The central argument is that AI-enhanced driving-behavior models should be evaluated not only by predictive accuracy, but also by their consistency with experimentally observed behavioral mechanisms and their ability to support interpretable, verifiable, and transferable mixed-traffic modeling.

Graphical abstract

Keywords

artificial intelligence / driving behavior / autonomous vehicles / connected mobility / federated learning / digital twin

CHANGE HISTORY

A Correction to this paper has been published:
https://doi.org

The Retraction Note was updated to reflect reception of statements of disagreement from all authors.

This article has been retracted.Please see the Retraction Notice for more detail:
https://doi.org

References

[1] Ahmed K I, Ben-Akiva M E, Koutsopoulos H N, Mishalani R G (1996). Models of freeway lane changing and gap acceptance behavior. In: Transportation and traffic theory, Proceedings of the 13th international symposium on transportation and traffic theory, lyon, france, July 1996, 501–515

[2] Alshiekh MBloem REhlers RKönighofer BNiekum STopcu U (2018). Safe reinforcement learning via shielding. In: Proceedings of the AAAI conference on artificial intelligence, 2669–2678

[3] Chellapandi V P, Yuan L, Brinton C G, Żak S H, Wang Z, (2024). Federated learning for connected and automated vehicles: A survey of existing approaches and challenges. IEEE Transactions on Intelligent Vehicles, 9( 1): 119–137

[4] Cui Y, Huang S, Zhong J, Liu Z, Wang Y, Sun C, Li B, Wang X, Khajepour A, (2024). Drivellm: Charting the path toward full autonomous driving with large language models. IEEE Transactions on Intelligent Vehicles, 9( 1): 1450–1464

[5] Dong J, Yang C, Cai M, Chen C, Xu Q, Wang J, Wang J, Li K, (2026). Multisource human-in-the-loop digital twin testbed for connected and autonomous vehicles in mixed traffic flow. Journal of Intelligent and Connected Vehicles, 9( 2): 9210084

[6] Feng S, Sun H, Yan X, Zhu H, Zou Z, Shen S, Liu H X, (2023). Dense reinforcement learning for safety validation of autonomous vehicles. Nature, 615( 7953): 620–627

[7] Feng S, Yan X, Sun H, Feng Y, Liu H X, (2021). Intelligent driving intelligence test for autonomous vehicles with naturalistic and adversarial environment. Nature Communications, 12( 1): 748

[8] Gazis D C, (2002). The origins of traffic theory. Operations Research, 50( 1): 69–77

[9] Gipps P G, (1986). A model for the structure of lane-changing decisions. Transportation Research Part B: Methodological, 20( 5): 403–414

[10] Guo J, Kurup U, Shah M, (2020). Is it safe to drive? An overview of factors, metrics, and datasets for driveability assessment in autonomous driving. IEEE Transactions on Intelligent Transportation Systems, 21( 8): 3135–3151

[11] Hu X, Li S, Huang T, Tang B, Huai R, Chen L, (2024). How simulation helps autonomous driving: A survey of Sim2real, digital twins, and parallel intelligence. IEEE Transactions on Intelligent Vehicles, 9( 1): 593–612

[12] Jiang R, Hu M, Zhang H M, Gao Z, Jia B, Wu Q S, Wang B, Yang M, (2014). Traffic experiment reveals the nature of car-following. PLoS One, 9( 4): e94351

[13] Jiang R, Hu M B, Zhang H M, Gao Z Y, Jia B, Wu Q S, (2015). On some experimental features of car-following behavior and how to model them. Transportation Research Part B: Methodological, 80: 338–354

[14] Kong X, Gao H, Shen G, Duan G, Das S K, (2022). Fedvcp: A federated-learning-based cooperative positioning scheme for social internet of vehicles. IEEE Transactions on Computational Social Systems, 9( 1): 197–206

[15] Koopman P, Wagner M, (2017). Autonomous vehicle safety: An interdisciplinary challenge. IEEE Intelligent Transportation Systems Magazine, 9( 1): 90–96

[16] Li T, Sahu A K, Talwalkar A, Smith V, (2020). Federated learning: Challenges, methods, and future directions. IEEE Signal Processing Magazine, 37( 3): 50–60

[17] Li XWang CLiu YHe DZhang JMa K (2026). PAVE: An end-to-end dataset for production autonomous vehicle evaluation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition 2026, 1010–1018

[18] Liao X, Zhao X, Wang Z, Zhao Z, Han K, Gupta R, Barth M J, Wu G, (2023). Driver digital twin for online prediction of personalized lane-change behavior. IEEE Internet of Things Journal, 10( 15): 13235–13246

[19] Mo Z, Shi R, Di X, (2021). A physics-informed deep learning paradigm for car-following models. Transportation Research Part C, Emerging Technologies, 130: 103240

[20] Myers D, Mohawesh R, Chellaboina V I, Sathvik A L, Venkatesh P, Ho Y H, Henshaw H, Alhawawreh M, Berdik D, Jararweh Y, (2024). Foundation and large language models: Fundamentals, challenges, opportunities, and social impacts. Cluster Computing, 27( 1): 1–26

[21] Peng M, Guo X, Chen X, Chen K, Zhu M, Chen L, Wang F Y, (2025). LC-LLM: Explainable lane-change intention and trajectory predictions with large language models. Communications in Transportation Research, 5: 100170

[22] Puiutta E, Veith E M S P, (2020). Explainable Reinforcement Learning: A Survey. Machine Learning and Knowledge Extraction, 77–95

[23] Rong R, Ma S, Ren N, Lin Q, Jia N, (2025). Generative artificial intelligence in intelligent transportation systems: A systematic review of applications. Frontiers of Engineering Management, 12( 4): 1020–1036

[24] Schölkopf B, Locatello F, Bauer S, Ke N R, Kalchbrenner N, Goyal A, Bengio Y, (2021). Toward causal representation learning. Proceedings of the IEEE, 109( 5): 612–634

[25] Shi H, Zhou Y, Wu K, Chen S, Ran B, Nie Q, (2023). Physics-informed deep reinforcement learning-based integrated two-dimensional car-following control strategy for connected automated vehicles. Knowledge-Based Systems, 269: 110485

[26] Shi Y, Wu T, Guo T, Huo J, Gu Z, Dai Y, Liu Z, (2025). Traffic simulation optimization considering driving styles. Communications in Transportation Research, 5: 100181

[27] Stern R E, Cui S, Delle Monache M L, Bhadani R, Bunting M, Churchill M, Hamilton N, Haulcy R, Pohlmann H, Wu F, Piccoli B, Seibold B, Sprinkle J, Work D B, (2018). Dissipation of stop-and-go waves via control of autonomous vehicles: Field experiments. Transportation Research Part C, Emerging Technologies, 89: 205–221

[28] Tian J, Jia B, Ma S, Zhu C, Jiang R, Ding Y, (2017). Cellular automaton model with dynamical 2D speed-gap relation. Transportation Science, 51( 3): 807–822

[29] Tian J, Jiang R, Jia B, Gao Z, Ma S, (2016a). Empirical analysis and simulation of the concave growth pattern of traffic oscillations. Transportation Research Part B: Methodological, 93: 338–354

[30] Tian J, Li G, Treiber M, Jiang R, Jia N, Ma S, (2016b). Cellular automaton model simulating spatiotemporal patterns, phase transitions and concave growth pattern of oscillations in traffic flow. Transportation Research Part B: Methodological, 93: 560–575

[31] Tian J, Zhang H M, Treiber M, Jiang R, Gao Z, Jia B, (2019). On the role of speed adaptation and spacing indifference in traffic instability: Evidence from car-following experiments and its stochastic model. Transportation Research Part B: Methodological, 129: 334–350

[32] Tian J, Zhu C, Chen D, Jiang R, Wang G, Gao Z, (2021). Car following behavioral stochasticity analysis and modeling: Perspective from wave travel time. Transportation Research Part B: Methodological, 143: 160–176

[33] Treiber M, Hennecke A, Helbing D, (2000). Congested traffic states in empirical observations and microscopic simulations. Physical Review E: Statistical Physics, Plasmas, Fluids, and Related Interdisciplinary Topics, 62( 2): 1805–1824

[34] Wang C, Jia D, Wang W, Ngoduy D, Peng B, Wang J, (2025). A knowledge-informed deep learning paradigm for generalizable and stability-optimized car-following models. Communications in Transportation Research, 5: 100211

[35] Wang Z, Han K, Tiwari P, (2022). Digital twin-assisted cooperative driving at non-signalized intersections. IEEE Transactions on Intelligent Vehicles, 7( 2): 198–209

[36] Xu Z, Zhang Y, Xie E, Zhao Z, Guo Y, Wong K Y K, Li Z, Zhao H, (2024). Drivegpt4: Interpretable end-to-end autonomous driving via large language model. IEEE Robotics and Automation Letters, 9( 10): 8186–8193

[37] Yang Q, Liu Y, Chen T, Tong Y, (2019). Federated Machine Learning: Concept and Applications. ACM Transactions on Intelligent Systems and Technology, 10( 2): 1–19

[38] You J, Gan R, Tang W, Huang Z, Liu J, Jiang Z, Shi H, Wu K, Long K, Fu S, Chen S, Ran B, (2025). FollowGen: A scaled noise conditional diffusion model for car-following trajectory prediction. Communications in Transportation Research, 5: 100215

[39] Yuan K, Huang Y, Yang S, Zhou Z, Wang Y, Cao D, Chen H, (2024). Evolutionary decision-making and planning for autonomous driving based on safe and rational exploration and exploitation. Engineering, 33: 108–120

[40] Zhang C, Tian C, Han T, Chen Y, Zhang J, Feng Y, (2026). Examining driving behaviors and trust in an in-vehicle warning system under uncertainty: A roundabout study. Journal of Intelligent and Connected Vehicles, 9( 1): 9210078

[41] Zhang R, Mao J, Wang H, Li B, Cheng X, Yang L, (2025). A survey on federated learning in intelligent transportation systems. IEEE Transactions on Intelligent Vehicles, 10( 5): 3043–3059

[42] Zhang S, Li J, Shi L, Ding M, Nguyen D C, Tan W, Weng J, Han Z, (2024). Federated learning in intelligent transportation systems: Recent applications and open problems. IEEE Transactions on Intelligent Transportation Systems, 25( 5): 3259–3285

[43] Zhao MRen BYang S (2026). Dangerous driving behavior detection method in vehicle intelligent cockpit based on transwindow multimodal fusion. Journal of Intelligent and Connected Vehicles

[44] Zheng S T, Jiang R, Tian J, Li X, Jia B, Gao Z, Yu S, (2023). A comparison study on the growth pattern of traffic oscillations in car-following experiments. Transportmetrica. B, Transport Dynamics, 11( 1): 706–724

[45] Zheng S T, Jiang R, Tian J, Li X, Treiber M, Li Z H, Gao L D, Jia B, (2022). Empirical and experimental study on the growth pattern of traffic oscillations upstream of fixed bottleneck and model test. Transportation Research Part C, Emerging Technologies, 140: 103729

[46] Zhou H, Ma K, Liang S, Li X, Qu X, (2024). A unified longitudinal trajectory dataset for automated vehicle. Scientific Data, 11( 1): 1123

[47] Zhou X, Liu M, Yurtsever E, Zagar B L, Zimmer W, Cao H, Knoll A C, (2025). Vision language models in autonomous driving: A survey and outlook. IEEE Transactions on Intelligent Vehicles, 1–20

Rights & permissions

Higher Education Press